Abstract: In the urban traffic network, the trip generations of every origin and trip attractions of every destination are closely related to the economic level of the traffic zone, land-use, population and the other factors, generally can be accurately forecasted, while the demands between the origin and the destination are difficult to accurately predict. In this paper, we assume that the trip generation of every origin and trip attraction of every destination is deterministic, but the demands between every origin and destination are uncertain and belong to a bounded interval. We propose a min-max model of continuous network design with demand uncertainty under user equilibrium with robust optimizations. For our proposed model, we apply sensitivity analysis combined with the methods of sequence average algorithm to solve the robust model. Numerical examples demonstrate that the robust solutions of the continuous network design are more reliable than the deterministic solutions.

Key words: systems engineering, continuous network design problem, demand uncertainty, robust optimization, mathematics programs with equilibrium constraint, sensitivity analysis

摘要： 在城市交通网络中，每个交通小区的交通发生量以及吸引量与其经济发展水平、土地利用、人口数量等因素密切相关，通常能够较为准确的预测，而小区之间的交通需求很难准确的预测. 本文假定所有起讫点的交通发生量和吸引量是确定的，而OD需求不确定且属于一个有界区间，利用鲁棒优化的方法建立OD需求不确定环境下考虑用户均衡约束的交通网络设计极小极大模型，并提出了灵敏度分析结合相继平均法(MSA)的求解算法. 数值算例表明，利用鲁棒优化方法得到的网络设计方案较传统的确定性网络设计方案具有更高的可靠性.

关键词: 系统工程, 连续交通网络设计问题, 需求不确定, 鲁棒优化, 带均衡约束的数学规划, 灵敏度分析